Nova Lite
AmazonAmazonProprietary
描述
A low-cost multimodal model that is lightning fast for processing images, video, documents, and text.
发布日期
2024-12-03
参数规模
—
上下文长度
300K
支持模态
image, pdf, text, video
能力雷达图
22
general
17
coding
22
reasoning
31
science
70
agents
90
multimodal
排行榜排名
基准测试分数 (LLM Stats)
(LLM Stats (zeroeval))Chat
IFEvalGoogle Research (2023)
89.7%自报
Finance
FinQA
73.6%自报
General
MMLU
80.5%自报
BFCL
66.6%自报
Language
Translation Set1→en COMET22
88.8%自报
Translation en→Set1 COMET22
88.8%自报
Translation Set1→en spBleu
43.1%自报
Translation en→Set1 spBleu
41.5%自报
Math
GSM8k
94.5%自报
MATH
73.3%自报
Multimodal
GroundUI-1K
80.2%自报
VATEX
77.8%自报
MM-Mind2Web
60.7%自报
MMMU
56.2%自报
Reasoning
ARC-C
92.4%自报
ChartQAMasry et al. (2022)
86.8%自报
HumanEvalOpenAI (2021)
85.4%自报
BBH
82.4%自报
DROP
80.2%自报
CRAG
43.8%自报
GPQANYU + Cohere + Anthropic (2023)
42.0%自报
Summarization
SQuALITY
19.2%自报
Vision
DocVQADocVQA (2020)
92.4%自报
TextVQA
80.2%自报
VisualWebBench
77.7%自报
EgoSchema
71.4%自报
LVBench
40.4%自报
AA 评测指数
(Artificial Analysis)Math 500(OpenAI (2024), subset of Hendrycks et al. MATH (2021))76.5
Mmlu Pro(TIGER-Lab (Univ. of Waterloo, Toronto, CMU, 2024))59.0
Gpqa(NYU + Cohere + Anthropic (2023))43.3
Ifbench(Google Research (2023))34.1
Lcr(Artificial Analysis)19.0
Tau2(Sierra + U Toronto + Vector Institute (2025))17.5
Livecodebench(UC Berkeley + MIT + Cornell (2024))16.7
Aime(MAA (Mathematical Association of America))10.7
Aime 25(MAA (Mathematical Association of America))7.0
Math Index(Artificial Analysis)7.0
Intelligence Index(Artificial Analysis)6.7
Hle(Center for AI Safety + Scale AI (2025))4.3
Terminalbench Hard(Stanford × Laude Institute (2026))0.8
LLM Stats 分类评分
(LLM Stats (zeroeval))Chat90
Image To Text90
Instruction Following90
Structured Output90
Code90
Legal80
Math80
Multimodal70
Reasoning70
Finance70
General70
Healthcare70
Tool Calling70
Vision70
Language60
Frontend Development60
Agents60
Economics60
Long Context40
Physics40
Search40
Biology40
Chemistry40
Summarization20
定价
输入价格$0.06 / 1M tokens
输出价格$0.24 / 1M tokens
混合价格(3:1)$0.105 / 1M tokens
缓存读取价格$0.01725 / 1M tokens
缓存写入价格$0.069 / 1M tokens
速度
Tokens/秒0.0
首Token延迟0.00s
首回答延迟0.00s
供应商价格排行
供应商价格排行
5 个供应商
最便宜: NanoGPT最贵: Amazon Bedrock
供应商输入输出
1NanoGPT最便宜
$0.0595
$0.238
2Amazon主要
$0.06
$0.24
3OpenRouter
$0.06
$0.24
4Kilo Gateway
$0.06
$0.24
5Amazon Bedrock
$0.069
$0.276
比较该模型在不同 API 供应商之间的定价。